Ego velocity assisted direction of arrival estimator for radar systems

Some radar systems are included in vehicles as part of automated driving assistance systems and used to assist in perception of environments around the vehicles. To accurately represent the environment, a radar system must determine the direction of arrival and motion state of objects in the environment. In addition, the radar system must determine the direction of arrival and motion state of object quickly enough to provide meaningful data to the automated driving assistance system. However, some techniques for determining the direction of arrival of objects in the environment are computationally expensive and slow to perform.

The use of the same reference symbols in different drawings indicates identical items unless otherwise noted. The Figures are not necessarily drawn to scale.

As discussed previously herein, some techniques for determining the direction of arrival and motion state of objects in the environment are computationally expensive and time-consuming. The disclosed techniques and radar systems implementing the disclosed techniques are able to determine a simplified two-dimensional (2D) angular search grid based on the ego velocity of the radar system and a range and radial velocity of the object in the environment. The simplified 2D angular search grid can dramatically reduce the number of search grid points to be analyzed, such that a conventional grid-based direction of arrival estimation algorithm can be performed much more quickly with a much lower computation cost compared to searching every azimuth and elevation in the field of view of the radar system. In addition, the simplified 2D angular search grid can be leveraged to provide motion state information for the object in the environment and to cross-check direction of arrival solutions to suppress false targets.

shows, in block diagram form, an example radar system, according to an embodiment. Radar systemcan be included in a vehicle, such as for an automotive driver assistance system. The example radar systemis a frequency-modulated continuous wave (FMCW) radar system, also referred to as a continuous-wave frequency-modulated (CWFM) radar, and capable of determining the distance or range, velocity, and angle of arrival of an object in the field of view of radar system. The term “angle of arrival” of an object is used herein to indicate the angle of arrival of a signal reflected off the object relative to the alignment of the radar system. Although radar systemis described herein as a linear chirp radar system, any appropriate radar system that generates range-Doppler information can be used.

In this example, radar systemincludes a microcontroller and processor unit (MCPU), a radar sensor circuit, a first antenna array, a second antenna array, and storage. The MCPUcomprises one or more MCPU cores, general purpose processing cores, array or vector processing cores, parallel processing cores, graphic processing units, neural net and linear algebra accelerators, field-programable gate arrays, digital signal processors, application-specific integrated circuits, and the like, or any combination thereof. The term “MCPU” in the singular is used herein to refer to either a single or multiple of the MCPU cores, general purpose processing cores, array or vector processing cores, parallel processing cores, graphic processing units, digital signal processors, neural net and linear algebra accelerators, application-specific integrated circuits, field-programable gate arrays, and the like, or any combination thereof comprised in the MCPU.

MCPUincludes a radar controllerand a signal processor. The radar controllercan receive data from the radar sensor circuitand control radar parameters of the radar sensor circuitsuch as frequency band, length of a radar frame, and the like via a serial peripheral interface (SPI), inter-integrated circuit (I2C) interface, an on-chip data bus and the like. A control signal from the radar controllercan be used to adjust the radar chirp signals output from a chirp generatorincluded in radar sensor circuit. The signal processorin MCPUcan also receive the data from the radar sensor circuitand perform signal processing for determining a distance or range between a target object and radar system, a radial velocity of the target object, an angle of arrival for the target object, and the like. The signal processorcan provide the calculated values to the storageand/or to other systems via the interface.

The interfacecan enable the MCPUto communicate with other systems over local and wide area networks, the internet, automotive communication buses, and/or other kinds of wired or wireless communication systems, for example. The MCPUcan provide the calculated values over the interfaceto other systems, such as a radar-camera-lidar fusion system; an automated driving assistance system including parking, braking, or lane-change assistance features; and the like. The storagecan be used to store instructions for the MCPU, received data from the radar sensor circuit, calculated values from the signal processor, and the like. Storagecan be any appropriate storage medium, such as a volatile or non-volatile memory.

The radar sensor circuitincludes the chirp generator, a transmitter, a receiver, a baseband processor, and an analog-to-digital converter (ADC). The chirp generatorcan include a local oscillator, for example, and generates radar chirp signals and provides them to the transmitter. For example, the chirp generatorfrequency can modulate a continuous wave signal to form a series of linear chirp signals. The transmitted chirp signal of a known, stable frequency continuous wave varies up and down in frequency over a fixed period of time by the modulated signal. The chirp generatorprovides the generated chirp signals to the transmitter, which drives the first antenna arrayof one or more transmitter (TX) antennas. The second antenna arraycomprises one or more receiver (RX) antennas and receives signals reflected from objects in the path of the transmitted chirp signals from the TX antenna array. The TX antenna arrayand the RX antenna arraycan be stationary or configured to transmit and receive across a range of area, such as by mechanical movement.

The receiverreceives the reflected signals from the RX antenna arrayand provides them to the baseband processor. The baseband processoralso receives the transmitted chirp signals from the chirp generatorand down-converts or chirp-compresses the received chirp signals directly into the baseband using the copy of the transmitted chirp signals from the chirp generator. The baseband processorcan then filter and amplify the baseband signal. The baseband processorprovides the filtered and amplified baseband signal to the ADC, which digitizes the signal and provides it to the MCPU. The signal processorin the MCPUcan then perform time domain to frequency domain transforms such as fast Fourier transforms (FFTs) and other signal processing to determine the distance, radial velocity, and angle of arrival between the target object and the radar system.

Frequency differences between the received reflections and the transmitted chirp signal increase with delay and so are proportional to distance. The phase differences between the received reflections across consecutive chirps in a radar frame are indicative of the velocity of objects in the field of view. For implementations in which RX antenna arrayincludes two or more receiver antennas, the phase difference between received reflections at a first RX antenna and received reflections at a second RX antenna can be used to determine the angle of arrival of target objects. For example, the down-converted and digitized receive signal corresponding to each chirp is first transformed using an FFT (called the range FFT). The range FFT produces a series of range bins with the value of each range bin denoting the signal strength of reflected targets at the corresponding range. A further “Doppler” FFT is then performed for each range bin across all the chirps in a frame to estimate the velocities of reflected targets. Additional processing can then be performed to determine the angle of arrival between the targets and the radar system. Although the radar systemis described herein as implementing FFT-based range-Doppler processing, any appropriate transforms may be used to produce the range-Doppler information.

For an implementation in which the radar systemis included in a vehicle with an automated driving assistance system, the automated driving assistance system can use the determined distance, velocity, and angle of arrival for objects in the field of view from the radar systemto provide parking, braking, or lane-change assistance. The radar systemmust determine the distance, velocity, and angle of arrival for objects in the field of view of the radar systemquickly enough to provide meaningful data to the automated driving assistance system. Radar systems according to the disclosed invention can determine the direction of arrival of objects in the environment more quickly than conventional techniques by using an ego velocity of the radar systems to reduce the search grid and the corresponding computational cost.

shows a diagramof a simplified 2D angular search grid for the example radar systemshown in, according to one embodiment. For ease of illustration, the diagramis described herein with reference to the radar systemshown in. The disclosed ego velocity assisted direction of arrival (EVADoA) algorithm leverages the ego velocity of the radar systemand the determined range and radial velocity of an object in the environment to determine a simplified 2D angular search grid that includes a ringand an arc. The radar systemhas an ego velocity Vegoat an angle relative to a forward direction; that is, relative to the forward direction, the radar systemmoves at an angle with an ego velocity Vego. The ego velocity Vegoof the radar systemcan be determined using a ground-reflection based ego velocity estimation algorithm or obtained from an external source, such as from a speedometer, a global positioning system (GPS), or an inertial measurement unit (IMU) of a vehicle on which the radar systemis included.

The radar systemdetermines that an object in the environment has a range Rfrom the radar systemand a radial velocity Vr, for example by performing a range FFT and a Doppler FFT on received radar data as described herein with respect to. The ringindicates possible positions of a stationary object in the environment, and the arcindicates possible positions of a moving object in the environment. The ringis vertical and centered on the direction of the ego velocity Vego, and the face of the ringis perpendicular to the direction of the ego velocity Vego. The opening angle between the ringand the direction of the ego velocity Vegois represented as an angle β, where the angle βis determined based on the ego velocity Vegoand the determined radial velocity Vr. For example, the radial velocity Vr can be represented as:

That is, each possible position in the ringis the distance Rfrom the radar system, centered around the direction of the ego velocity Vego. The angle between the ego velocity Vegoand the velocity of the object represented as cos(β)*Vis the angle β. A width of the ringcan be chosen based on the resolution of the determined radial velocity, such that each possible position in the ringhas a projected velocity approximately equal to the determined radial velocity, within a tolerance determined by the resolution of the determined radial velocity. For example, a tolerance e can be chosen based on the sum of the radial velocity measurement resolution and error and the ego velocity measurement resolution and error, such that:

Possible positions in the ringthat are outside a relevant elevation field of view can be omitted. For example, a radar systemhas an elevation field of view of −30° to 30°; any possible positions in the ringwith elevations less than −30° or greater than 30° can be omitted.

The arcindicates possible positions of a moving object in the environment. In implementations in which the radar systemis included in a vehicle as part of an automated driving assistance system, moving objects of interest to the automated driving assistance system such as pedestrians, strollers, bicycles, other vehicles, and the like are present on the ground surface. That is, moving objects that are in the air, such as airplanes or birds, or that are under the ground, such as subway cars, are not relevant or of interest to the automated driving assistance system. As a result, the search grid for moving objects can be reduced to the arcof possible positions that are on the ground surface and the distance Rfrom the radar system. In some implementations, the arcof possible positions of a moving object in the environment can be at a 0° elevation. In other implementations, the arcof possible positions of a moving object in the environment can be chosen based on a maximum relevant height and the range Rfrom the radar system, as described further herein with respect to.

The disclosed EVADoA algorithm simplifies the 2D azimuth-elevation search grid for determining the direction of arrival of objects in the environment to the ringand the arc. For example, a radar systems has an azimuth field of view of −90° to 90° with a 0.5° grid size and an elevation field of view of −30° to 30° with a 0.5° grid size. The total 2D search grid for the direction of arrival is 361 by 121, totaling 43,681 search points. The simplified 2D search grid includes the ringand the arc. The ringcan include at most 361 search points, if the ringis entirely within the elevation field of view, and fewer than 361 search points if the ringis partially cut off into two segments. The arccan be a single azimuth search of 361 search points at an elevation angle corresponding to the ground surface. Thus, the simplified 2D search grid determined by the disclosed EVADoA algorithm can result in a 98% reduction in the computational cost for determining the direction of arrival.

For a known range R, radial velocity, and Vego, as well as the azimuth and elevation fields of view, a directional vectorfrom the phase center of the radar systemto an i-th angular search point (φ, θ) can be represented as:

where θis the elevation angle in radians for the i-th search point, φis the azimuth angle in radians for the i-th search point,is a unit-length directional vector perpendicular to the front directionand pointing to the right (i.e., pointing 90 degrees clockwise from the front directionon the horizontal plane),is a unit-length direction vector in the front direction, andis a unit-length direction vector perpendicular upwards of front directionand normal to the horizontal plane, and assuming the range is equal to the known range Rwithin a range tolerance b. Every point on the arcand the ringresults in the same range Rto the phase center of the radar. In some implementations, the range tolerance bis representative of a 95% confidence error bound for the range R.

The directional vectorfor the i-th search point (φ, θ) and the ego velocity Vegoare related by the angle β, which can be represented as:

where “·” represents an inner product operation. For the i-th search point (φ, θ) to qualify as a possible position of a stationary object—that is, for the i-th search point (φ, θ) to be included in the ring—the elevation angle θmust be within the elevation field of view, and the azimuth angle φmust be within the azimuth field of view. In addition, the i-th search point (φ, θ) qualifies as a possible position of a stationary object in response to the radial velocity Vr, the angle β, and the ego velocity Vegosatisfying the condition represented as:

where ϵ represents a tolerance threshold determined based on the radial velocity measurement resolution and error and the ego velocity measurement resolution and error. For example, the tolerance threshold ϵ can be represented as:

where brepresents an ego velocity error tolerance and brepresents a radial velocity error tolerance. In some implementations, band bare representative of 95% confidence error bounds.

For the i-th search point (φ, θ) to qualify as a possible position of a moving object—that is, for the i-th search point (φ, θ) to be included in the arc—the azimuth angle φmust be within the azimuth field of view. In addition, the elevation angle θmust be at least within the elevation field of view. The height of the radar systemand a maximum relevant object height can further reduce the number of possible elevation angles for the elevation angle θ, as described further herein with respect to.

shows a diagramof a relevant elevation field of view for the example radar system shown in, according to one embodiment. For ease of illustration, the diagramis described herein with respect to the radar systemshown in. The diagramincludes the radar systemhaving a radar heightabove the ground surfaceand a moving objectthat has an object heightand is a distance Rfrom the radar system. The relevant elevation field of viewof the radar systemcan be chosen to extend from the ground surfaceto a maximum relevant object height. For example, in implementations in which the radar systemis incorporated into a vehicle as part of an automated driving assistance system, the maximum relevant object height can be chosen to be a maximum height of semi-trailer trucks on the road, such as 14 feet. The object heightis less than the maximum height of the relevant elevation field of view and greater than 0.

As discussed previously herein, the height of the radar systemand the maximum relevant object height can further reduce the number of possible elevation angles for the elevation angle θin the i-th search point (φ, θ), for the i-th search point (φ, θ) to be included in the arc. An object's elevation angle θ is related to the height of the radar systemand the maximum relevant object height by the equation:

where Hmax represents the maximum relevant object height, Hradar is the height of the radar system, and R represents the range R. Thus to be included in the arc, the elevation angle θfor the i-th search point (φ, θ) must be greater than or equal to the larger of the minimum elevation field of view and

and less than or equal to the smaller of the maximum elevation field of view and

shows, in flow chart form, an example processfor determining directions-of-arrival for objects in an environment using a simplified two-dimensional angular search grid, according to one embodiment. For ease of illustration, the processis described herein with reference to the radar systemshown inand the diagramshown in. The processis described herein as being performed by the signal processorin the MCPUexecuting instructions stored in storageof the radar system, but can be performed by any appropriate processing unit executing instructions stored in any appropriate non-transitory computer-readable medium. The steps of processare shown in a particular order in, but the steps of processmay be performed in a different order and/or some steps may be performed simultaneously. One or more steps of processcan be optional, and processcan include additional steps.

The process begins at step, at which signal processorobtains an ego velocity Vegoof the radar systemand a range Rand a radial velocity Vr of an object in an environment of the radar system. For example, the ego velocity Vegoof the radar systemcan be determined using a ground-reflection based ego velocity estimation algorithm or obtained from an external source, such as from a speedometer or GPS/IMU module of a vehicle on which the radar systemis included. The range Rand the radial velocity Vr can be obtained, for example, by performing a range FFT and a Doppler FFT on received radar data as described herein with respect to. At step, the signal processordetermines a simplified 2D angular search grid.

Stepincludes stepsand. At step, the signal processordetermines a ringof possible positions for a stationary object in the environment based on the ego velocity Vego, the range R, and the radial velocity Vr. For example as described further herein with respect to, an i-th search point (φ, θ) is included in the ringin response to the elevation angle θbeing within the elevation field of view, the azimuth angle φbeing within the azimuth field of view, and the radial velocity Vr, the angle β, and the ego velocity Vegosatisfying the condition represented as:

At step, the signal processordetermines an arcof possible positions for a moving object in the environment based on the range Rof the object. For example, the arccan be at an elevation corresponding to a ground surface of the environment and the possible positions having the range Rfrom the radar system. In some implementations, the arccan be chosen based on the height Hradar of the radar systemand the maximum relevant object height Hmax. That is, the arccan include possible positions having elevations within the elevation field of view, greater than or equal to:

and less than or equal to

At step, the signal processorperforms a grid-based direction-of-arrival estimation algorithm using the simplified 2D angular search grid determined in step. The signal processorcan use any appropriate grid-based direction-of-arrival estimation algorithm, such as a Fourier beamformer (e.g. Fast Fourier Transform or Discrete Fourier Transform), a Capon beamformer, an orthogonal matching pursuit (OMP) algorithm, a multiple signal classification (MUSIC) algorithm, deterministic maximum likelihood (DML) algorithm, a Basis Pursuit (BP), an Iterative Adaptive Approach (IAA), and the like. The simplified 2D angular search grid including the ringand the arcreduces the computational cost of performing the grid-based direction-of-arrival estimation algorithm at stepcompared to techniques that scan the entire azimuth and elevation fields of view. The result of the processis a direction of arrival for the object in the environment, as well as motion state information, such as whether the object is moving or stationary.

shows, in flow chart form, an example processfor determining directions-of-arrival for objects in an environment using a simplified elevation angular search grid, according to one embodiment. For ease of illustration, the processis described herein with reference to the radar systemshown inand the diagramshown in. The processis described herein as being performed by the signal processorin the MCPUexecuting instructions stored in storageof the radar system, but can be performed by any appropriate processing unit executing instructions stored in any appropriate non-transitory computer-readable medium. The steps of processare shown in a particular order in, but the steps of processmay be performed in a different order and/or some steps may be performed simultaneously. One or more steps of processcan be optional, and processcan include additional steps.

The process begins at step, at which signal processorobtains an ego velocity Vegoof the radar systemand a range Rand a radial velocity Vr of an object in an environment of the radar system. For example, the ego velocity Vegoof the radar systemcan be determined using a ground-based ego velocity estimation algorithm or obtained from an external source, such as from a speedometer or GPS/IMU module of a vehicle on which the radar systemis included. The range Rand the radial velocity Vr can be obtained, for example, by performing a range FFT and a Doppler FFT on received radar data as described herein with respect to. At step, the signal processorperforms an azimuth direction-of-arrival estimation algorithm. Any appropriate azimuth direction-of-arrival estimation algorithm can be used, such as a Fourier beamformer (e.g. Fast Fourier Transform or Discrete Fourier Transform), a Capon beamformer, an OMP algorithm, a MUSIC algorithm, a DML algorithm, a BP algorithm, an IAA, and the like.

At step, the signal processordetermines a simplified 2D angular search grid based on the azimuth of the object determined at step. Stepincludes steps,, and. At step, the signal processordetermines a ringof possible positions for a stationary object based on the ego velocity Vego, the range R, and the radial velocity Vr. For example, an i-th search point (φ, θ) is included in the ringin response to the elevation angle θbeing within the elevation field of view, the azimuth angle φbeing within the azimuth field of view, and the radial velocity Vr, the angle β, and the ego velocity Vegosatisfying the condition represented as:

At step, the signal processordetermines an arcof possible positions for a moving object in the environment based on the range Rof the object. For example, the arccan be at an elevation corresponding to a ground surface of the environment and include the possible positions having the range Rfrom the radar system. In some implementations, the arccan be chosen based on the height Hradar of the radar systemand the maximum relevant object height Hmax. That is, the arccan include possible positions having elevations within the elevation field of view, greater than or equal to

and less than or equal to

At step, the signal processorselects relevant possible positions from the ringand the arcbased on the estimated azimuth. That is, the signal processorselects the possible positions in the ringand the arcthat have the estimated azimuth, such that the simplified 2D angular search grid includes the portions of the ringhaving the estimated azimuth and the portion of the arcthat has the estimated azimuth. At step, the signal processorperforms a grid-based elevation direction-of-arrival estimation algorithm using the simplified 2D angular search grid determined in step. The signal processorcan use any appropriate grid-based elevation direction-of-arrival estimation algorithm, such as a Fourier beamformer (e.g. Fast Fourier Transform or Discrete Fourier Transform), a Capon beamformer, an OMP algorithm, a MUSIC algorithm, a DML algorithm, a BP algorithm, an IAA, and the like. The simplified 2D angular search grid determined at stepcan reduce the computational cost of determining the elevation of the object compared to techniques that scan the entire elevation field of view at the determined azimuth of the object. The result of the processis a direction of arrival for the object in the environment, as well as motion state information, such as whether the object is moving or stationary.

shows, in flow chart form, an example processfor determining a direction-of-arrival and motion state for an object in an environment, according to one embodiment. For ease of illustration, the processis described herein with reference to the radar systemshown inand the diagramshown in. The processis described herein as being performed by the signal processorin the MCPUexecuting instructions stored in storageof the radar system, but can be performed by any appropriate processing unit executing instructions stored in any appropriate non-transitory computer-readable medium. The steps of processare shown in a particular order in, but the steps of processmay be performed in a different order and/or some steps may be performed simultaneously. One or more steps of processcan be optional, and processcan include additional steps.

The process begins at step, at which signal processorobtains an ego velocity Vegoof the radar systemand a range Rand a radial velocity Vr of an object in an environment of the radar system. For example, the ego velocity Vegoof the radar systemcan be determined using a ground-reflection based ego velocity estimation algorithm or obtained from an external source, such as from a speedometer or GPS/IMU module of a vehicle on which the radar systemis included. The range Rand the radial velocity Vr can be obtained, for example, by performing a range FFT and a Doppler FFT on received radar data as described herein with respect to. At step, the signal processorobtains an estimated target azimuth and elevation (φ, θ) for the object. For example, the signal processorcan obtain the estimated target azimuth and elevation (φ, θ) for the object from another processor.

At step, the signal processorperforms an EVADoA validation process, such as the EVADoA validation processdescribed further herein with respect to. The outputfrom the EVADoA validation process performed at stepis a target direction-of-arrival validity flag and motion state information for a valid direction-of-arrival. That is, the outputincludes a target direction-of-arrival validity flag that is indicative of whether the estimated target azimuth and elevation (φ, θ) for the object is valid, and if so, motion state information for the object, such as whether the object is stationary or moving.

shows, in flow chart form, an example EVADoA validation processfor checking an estimated target azimuth and elevation (φ, θ) and motion state for an object using a simplified two-dimensional angular search grid, according to one embodiment. For ease of illustration, the processis described herein with reference to the radar systemshown inand the diagramshown in. The processis described herein as being performed by the signal processorin the MCPUexecuting instructions stored in storageof the radar system, but can be performed by any appropriate processing unit executing instructions stored in any appropriate non-transitory computer-readable medium. The steps of processare shown in a particular order in, but the steps of processmay be performed in a different order and/or some steps may be performed simultaneously. One or more steps of processcan be optional, and processcan include additional steps.

The process begins at step, at which signal processordetermines whether the estimated azimuth φis within the minimum and maximum azimuth field of view of the radar system. In response to the estimated azimuth φnot being within the minimum and maximum azimuth field of view, the signal processoroutputs an indicationof an invalid estimated target azimuth and elevation. In response to the estimated azimuth φbeing within the minimum and maximum azimuth field of view, the signal processorproceeds to step, and determines whether the estimated elevation θis within the minimum and maximum elevation field of view of the radar system. In response to the estimated elevation θnot being within the minimum and maximum elevation field of view, the signal processoroutputs an indicationof an invalid estimated target azimuth and elevation.